Statistical analysis of speech is an emerging area of machine learning. In this paper, we tackle the biometric challenge of Automatic Speaker Verification (ASV) of differentiating between samples generated by two distinct populations of utterances, those of an authentic human voice and those generated by a synthetic one. Solving such an issue through a statistical perspective foresees the definition of a decision rule function and a learning procedure to identify the optimal classifier. Classical state-of-the-art countermeasures rely on strong assumptions such as stationarity or local-stationarity of speech that may be atypical to encounter in practice. We explore in this regard a robust non-linear and non-stationary signal decomposition method known as the Empirical Mode Decomposition combined with the Mel-Frequency Cepstral Coefficients in a novel fashion with a refined classifier technique known as multi-kernel Support Vector machine. We undertake significant real data case studies covering multiple ASV systems using different datasets, including the ASVSpoof 2019 challenge database. The obtained results overwhelmingly demonstrate the significance of our feature extraction and classifier approach versus existing conventional methods in reducing the threat of cyber-attack perpetrated by synthetic voice replication seeking unauthorised access.

Increased absenteeism in health care institutions is a major problem, both economically and health related. Our objectives were to understand the general evolution of absenteeism in a university hospital from 2007 to 2019 and to analyze the professional and sociodemographic factors influencing this issue. An initial exploratory analysis was performed to understand the factors that most influence absences. The data were then transformed into time series to analyze the evolution of absences over time. We performed a temporal principal components analysis (PCA) of the absence proportions to group the factors. We then created profiles with contributions from each variable. We could then observe the curves of these profiles globally but also compare the profiles by period. Finally, a predictive analysis was performed on the data using a VAR model. Over the 13 years of follow-up, there were 1,729,097 absences for 14,443 different workers (73.8% women; 74.6% caregivers). Overall, the number of absences increased logarithmically. The variables contributing most to the typical profile of the highest proportions of absences were having a youngest child between 4 and 10 years old (6.44% of contribution), being aged between 40 and 50 years old (5.47%), being aged between 30 and 40 years old (5.32%), working in the administrative field (4.88%), being tenured (4.87%), being a parent (4.85%), being in a coupled relationship (4.69%), having a child over the age of 11 (4.36%), and being separated (4.29%). The forecasts predict a stagnation in the proportion of absences for the profiles of the most absent factors over the next 5 years including annual peaks. During this study, we looked at the sociodemographic and occupational factors that led to high levels of absenteeism. Being aware of these factors allows health companies to act to reduce absenteeism, which represents real financial and public health threats for hospitals.

A practical algorithm has been developed for closeness analysis of sequential data that combines closeness testing with algorithms based on the Markov chain tester. It was applied to reported sequential data for COVID-19 to analyze the evolution of COVID-19 during a certain time period (week, month, etc.).

Agro-climatic indicators (AgcIs) provide a suitable tool to evaluate the implications of climate change on agriculture by simplifying plant–climate interactions. However, developing reliable AgcIs requires high-quality historical climate datasets. Consequently, reanalysis products (RPs) are frequently used as a potential reference dataset for observed climate in agricultural studies. This study aims to compare five RPs (ERA5, ERA5-Land, SCOPE Climate, FYRE Climate, and RFHR) at reproducing observed AgcIs over France. The RPs are evaluated against the SYNOP meteorological data over the 1996–2012 period, focusing on six AgcIs specific to apple, maize, and vine crops. The findings show that RPs perform well in reproducing temperature-based AgcIs, with some slight discrepancies in areas with complex topography. However, all RPs tend to overestimate precipitation amounts and to underestimate dry days, leading to a poor performance in reproducing precipitation-based AgcIs. This study emphasizes the need for a thorough evaluation of the RPs in developing both temperature-based and precipitation-based AgcIs, especially if findings are intended to support operational agricultural decision-making.